Excess upper-body adiposity, particularly in the visceral region, confers increased risk of developing metabolic syndrome, type 2 diabetes and cardiovascular disease (Neeland et al., 2012; Abraham et al., 2015). Since neck fat compartments expand with increasing adiposity, interest has focused on the utility of neck circumference (NC), a simple anthropometric measurement and proxy for upper-body subcutaneous fat, to predict cardiometabolic risk (Torriani et al., 2014). Imaging studies have established that neck adipose tissue quantified by computed tomography correlates positively with abdominal visceral fat and cardiovascular risk factors, and is predictive of metabolic syndrome and obstructive sleep apnea severity (Karimi et al., 2010; Li et al., 2014; Torriani et al., 2014). In large population studies, such as the Framingham Heart Study and the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), NC was positively associated with cardiovascular risk factors and subclinical atherosclerosis, even after adjustment for visceral adipose tissue and/or body mass index (BMI) (Preis et al., 2010; Baena et al., 2016). This suggests that fat accumulation in the neck may confer risk beyond visceral adipose tissue accumulation. Moreover, in longitudinal analyses, NC was predictive of incident type 2 diabetes in the Framingham cohort, and of future cardiovascular events and all-cause mortality in high-risk cardiology outpatients (Preis et al., 2013; Dai et al., 2016).

Elevated sympathetic nervous system (SNS) activity is a characteristic of obesity that relates to a number of pathogenic factors including central body fat distribution, hyperinsulinemia secondary to insulin resistance, baroreflex impairment, and the presence of obstructive sleep apnea (OSA) (Narkiewicz et al., 1998a; Beske et al., 2002; Grassi et al., 2004). Chronic sympathetic activation promotes unfavorable alterations in cardiovascular structure and function (e.g., arterial stiffening, cardiac enlargement, diastolic dysfunction, glomerular hyperfiltration) and decreased insulin sensitivity, via both blood pressure dependent and independent effects. Consequently, it is an important mediator of target organ damage and metabolic risk in both obese and hypertensive states (Masuo et al., 1997; Schlaich et al., 2015). To our knowledge, no study to date has examined the relationship of NC to SNS activity, which may be of relevance to cardiovascular risk stratification. Therefore, the aim of the present study was to perform gender-specific cross-sectional analyses to determine whether NC is independently associated with muscle sympathetic neural activity (MSNA) in overweight and obese subjects. The technique of microneurography was used to directly quantify efferent post-ganglionic sympathetic outflow directed to skeletal muscle vasculature in the lower leg.

Of the 125 subjects studied, 72 were men and 53 were women. The two genders had similar BMI, HOMA-IR, Matsuda ISI, glucose tolerance, MSNA and cardiac BRS. However, the women were significantly older (58 ± 1 vs. 54 ± 1 years), with higher plasma HDL- and LDL-cholesterol, leptin and NEFA concentrations, and lower alanine aminotransferase (ALT) and gamma-glutamyltransferase (GGT) levels, body weight, waist circumference, waist to hip ratio and NC compared to the men. (P all < 0.01).

Tables 1, 2 summarize clinical and demographic data according to NC tertiles in men and women, respectively. Men within the highest tertile of NC had significantly higher anthropometric indices, fasting and post-glucose plasma insulin levels (insulin area under the curve, AUC₀₋₁₂₀), HOMA-IR, NEFA, and lower Matsuda ISI compared to men in the lowest tertile (P all < 0.05). Blood pressure and cardiac BRS did not differ across NC tertiles. In women, BMI, waist circumference and waist to hip ratio increased significantly across NC groups, whereas metabolic and cardiovascular parameters did not differ between tertiles. Figure 1 shows that men within the highest NC tertile had significantly higher MSNA burst frequency (45 ± 2 vs. 36 ± 2 bursts/min) and burst incidence (69 ± 3 vs. 58 ± 3 bursts/100 hb) compared to the lowest NC tertile. In women MSNA did not differ between NC tertiles (lower panels, Figure 1).

Table 3 summarizes univariate correlates of MSNA and NC. In men, NC and insulin AUC₀₋₁₂₀ were the strongest positive correlates of MSNA. In women, age and plasma LDL-cholesterol concentration were most closely associated with MSNA. In adjusted stepwise regression analyses, NC alone explained 12% of the variance in MSNA in men. Together with insulin AUC₀₋₁₂₀, NC explained 22% of the variability in men (Table 4). In women, burst frequency was independently associated with age and HOMA-IR, which explained 19% of the variance, whilst burst incidence was independently associated with LDL-cholesterol and HOMA-IR, together explaining 18% of the variance (P < 0.05).

The present study investigated the association of NC, a potential marker for upper body fat deposition, with resting sympathetic neural activity in un-medicated middle-aged persons who were overweight or obese. Findings indicated a dichotomy of effects within the two genders. In men, NC was positively associated with MSNA independent of age, blood pressure and other adiposity measures such as waist circumference and BMI. In women, NC was associated with anthropometric measures, but did not delineate MSNA, metabolic or cardiovascular risk factors. Our data suggest that hyperinsulinemia secondary to decreased insulin sensitivity, gauged as elevated fasting and post-glucose insulin AUC₀₋₁₂₀, may be an intermediary mechanism underlying the association between NC and MSNA in men. NC was not associated with spontaneous cardiac BRS in either gender in the present study.

Our findings concur with previously published data in Caucasian and Asian populations, showing an association of NC with classic obesity indices such as waist circumference and BMI, and with insulin-resistance related factors (Laakso et al., 2002; Zhou et al., 2013). More recent imaging studies have highlighted that NC is positively correlated with visceral adipose tissue and epicardial fat (Li et al., 2014; Küç¨uk et al., 2016) and with subclinical and clinical atherosclerosis (Zen et al., 2012; Baena et al., 2016). Our data further extend these observations, by demonstrating for the first time that NC is correlated with MSNA, a direct measure of central sympathetic neural outflow to skeletal muscle, in men. Various pathogenic mechanisms may be responsible for this association. Firstly, hyperinsulinemia, which related to both NC and MSNA, is known to elicit sympathetic activation via central and peripheral mechanisms (Anderson et al., 1991; Kern et al., 2005). Centrally, insulin's effects are initiated in the arcuate nucleus and involve various neural signaling pathways that mediate organ-specific sympathetic outflows (Rahmouni et al., 2004; Cassaglia et al., 2011). Peripherally, insulin may evoke a reflex increase in SNS activity in response to its vasodilatory effects, albeit insulin-induced sympathetic activation and vasodilation in skeletal muscle are known to be blunted in obese humans (Vollenweider et al., 1994). Reverse causality may also be active, in that chronic sympathetic activation per se may worsen insulin resistance via hemodynamic and lipolytic effects, thus maintaining a vicious bidirectional cycle (Jamerson et al., 1993; Masuo et al., 1997). Interestingly, increasing adiposity was not associated with worsening insulin sensitivity in the women participants of our study.

Secondly, greater NC is associated with OSA syndromes (Karimi et al., 2010; Lim et al., 2014) which are known to be an important mediator of elevated sympathetic drive in obesity. This is attributed to the stimulation of peripheral chemoreceptors by the anoxia induced by nocturnal hypopneic-apnoeic episodes, and also to diminished arterial baroreflex sensitivity in OSA subjects (Narkiewicz et al., 1998b; Toschi-Dias et al., 2013). The prevalence of OSA in newly diagnosed metabolic syndrome subjects with similar clinical profile to our cohort, has been reported at 60% (Drager et al., 2010). In the absence of formal polysomnographic testing, we can only speculate about the potential involvement of OSA in the association between NC and MSNA in the present study. Notably, NC is more strongly associated with OSA in men than women, which may be one potential explanation for the gender differences observed in our study (Lim et al., 2014).

Whilst no study to date has examined the relation of NC to baroreflex function, it is recognized that visceral obesity and the metabolic syndrome are associated with impairment in baroreceptor ability to control heart rate and efferent postganglionic sympathetic nerve traffic (Beske et al., 2002; Grassi et al., 2005). Attenuation of these reflex mechanisms stemming from carotid and aortic baroreceptors is thought to play a causal role in the hyperadrenergic state typical of the metabolic syndrome. In the present study we found no relation between NC and spontaneous cardiac BRS, which primarily reflects vagal efferent activity, in either men or women. However, it merits emphasis that we did not analyse baroreceptor modulation of MSNA, which may be of particular relevance. Notably, in patients with OSA, impairment of baroreflex sympathetic modulation is not accompanied by any impairment in baroreflex control of heart rate (Narkiewicz et al., 1998b).

Dissimilarities in neck fat accumulation have been demonstrated in men and women using positron emission and computed tomography imaging. Overweight and obese women have significantly higher neck subcutaneous fat and significantly lower neck intermuscular adipose tissue compared to age- and BMI-matched men (Torriani et al., 2014). Posterior intermuscular and subcutaneous neck adipose tissue depots have been linked with cardiometabolic risk (Torriani et al., 2014). Moreover, NC is more closely related to abdominal visceral fat in men than women (Li et al., 2014). A number of previous studies, including our own work, support the notion that gender influences the interaction between adiposity indices and SNS activity. BMI and waist circumference are closely associated with sympathetic nerve traffic in men but not women (Lambert et al., 2007; Tank et al., 2008). The biological basis for these differences is yet to be elucidated (Brooks et al., 2015). There are recognized sex differences in the distribution, morphology and activity of adipose tissue. Adipose tissue generates various signals that may modulate sympathetic activity, including leptin, angiotensinogen and NEFA, and there may be gender-specific differences in the quantities produced and actions elicited on sympathetic pathways (Brooks et al., 2015). For example, rodent studies show that leptin's sympathoexcitatory effects are exaggerated in obese male animals, whereas in ovariactomized female rats, intracerebroventricular leptin failed to alter lumbar sympathetic nerve activity, unless the rats were given estrogen replacement therapy (Shi and Brooks, 2015). In diet-induced obese mice, weight gain was associated with increased plasma angiotensin II in male, but not female animals (Gupte et al., 2012). Blunted leptin-mediated sympathetic activation, quantified as whole-body norepinephrine spillover rate, has also been linked to leptin receptor polymorphisms, highlighting that genetic factors may modulate the relation between adiposity hormones and SNS activity (Masuo et al., 2008). As far as the present study is concerned, we found positive associations between NEFA, MSNA and NC, and leptin and NC in the men, but no further relationships with other adiposity signals. Insulin resistance is the primary mechanisms leading to greater free fatty acid release in upper body obesity. Enhanced lipolysis produced by activation of the SNS is another contributing factor. In turn, NEFA promote insulin resistance via the Randle cycle and may thus increase SNS activity indirectly (via hyperinsulinemia). Moreover, lipid administration studies suggest that NEFA may also have direct effects on α1-adrenoceptor sensitivity and central sympathetic drive (Stepniakowski et al., 1995; Florian and Pawelczyk, 2010).

The strengths of our study include the use of un-medicated subjects to avoid confounding by pharmacological agents. Furthermore, microneurographic recordings are highly reproducible within individuals and provide a direct measure of postganglionic efferent nerve discharges leading to skeletal muscle vasculature, which are of particular relevance to vascular tone and glucose utilization (Fagius and Wallin, 1993; Jamerson et al., 1993; Fairfax et al., 2013; Gamboa et al., 2014). Our study also has several limitations. First, NC is a measure that comprises both adipose and lean tissue and our study does not distinguish between these compartments. Second, the sample size was relatively small and we did not include any lean controls, which would have enhanced the heterogeneity of the cohort, and possibly the strength of associations. Third, a number of unmeasured factors such as OSA, fitness level and androgenicity may have influenced the relationships that were studied. Fourth, we only examined sympathetic outflow to skeletal muscle but not to other organs such as the kidneys, which are of relevance to the development of hypertension and cardiovascular risk. Finally, the cross-sectional design does not permit insight into causality.

In summary, within overweight and obese men NC was associated with MSNA and features of insulin resistance independent of central adiposity. In women NC was associated with anthropometric measures but not with MSNA or metabolic indices. Given the clinical relevance of chronic sympathetic overactivity to future cardiovascular and metabolic disease (Masuo et al., 1997; Schlaich et al., 2015; Wulsin et al., 2015), NC may be a simple measure to aid risk stratification in at risk overweight and obese men. Our data does not support its use in women, beyond identifying overweight and obesity.

NS and EL conceived the study. NS, MG, CS, PN, and MS collected clinical data. EL performed microneurographic recordings and analyses. NE and SP performed laboratory assays. NS performed statistical analyses, wrote and revised the manuscript. NS, PN, JD, GL, and EL interpreted the data. All authors critically read and approved the manuscript.

This study was funded by a Diabetes Australia Millennium Grant in type 2 diabetes to NS. JD, GL, and MS are supported by NHMRC Fellowships. We also wish to acknowledge the Victorian Government's Operational Infrastructure Support Program.